The purpose of the proposed predoctoral and postdoctoral programs is to train the next generation of cognitive psychologists both to develop formal computational models and to test and refine these models by rigorously comparing the simulation data to carefully collected empirical data. The field is ready to benefit from formal, computational models of cognitive processes. The tools are being developed that enable this formalism, and the end product will not only deepen the empirical and conceptual basis of cognitive psychology, but will also provide stronger links between psychology, neuroscience, and the treatment of problems in mental health. Carnegie Mellon is especially suited to provide this next generation of cognitive scientists with these tools. There is a long tradition at CMU to strive for complete cognitive models to account for a wide range of phenomena using a small common set of theoretical assumptions. The proposed program would be our first focused on training modeling skills. One of the distinctive features of psychological research at CMU is the dual concern for experimental methodology and theoretical models, not just each in isolation. We have promoted the development of both production system (symbolic), connectionist (sub-symbolic) and hybrid models of the human information processing architecture as well as many specific models of performance in particular tasks. In all cases, the researchers have tested and refined their models based on behavioral and physiological data collected here at CMU and elsewhere. Methodologies that have been developed and refined within our department include: the automatic coding of verbal protocols, the analysis of eye fixations while thinking and problem solving, and functional MRI measurements of higher cognitive processes. Some of these models address the data at the grain size of individual responses, with few subject-specific parameters. The program's goal is to develop skilled researchers who are both competent and comfortable combining the approaches of behavioral research with development of computationally implemented models of cognitive performance. Participation in research, both empirical and modeling, is a fundamental component of helping students achieve this goal. Formal courses and seminars will play in important role as well. We will formally instruct and demonstrate the skills of comparing the data derived from a simulation to the human data collected from behavioral research, providing students with the skills to evaluate the quality of the fit and the sensitivity to know when and how to revise one's model based on these comparisons. In addition we will insure that trainees are conversant in multiple computational approaches and recognize the strengths and weaknesses of each.